Add new unalign flag to README.md and update help text

This commit is contained in:
David Addison 2026-03-06 17:45:03 -08:00
parent e986a6156c
commit 115fb09377
2 changed files with 4 additions and 3 deletions

View File

@ -83,7 +83,8 @@ All tests support the same set of arguments :
* `-S,--report_timestamps <0/1>` Add timestamp (`"%Y-%m-%d %H:%M:%S"`) to each performance report line. Default : 0.
* `-J,--output_file <file>` Write [JSON] output to filepath. Infer type from suffix (only `json` supported presently).
* `-T,--timeout <time in seconds>` timeout each test after specified number of seconds. Default : disabled.
* `-M,--memory_report <0/1>` enable memory usage report. Default : 0
* `-M,--memory_report <0/1>` enable memory usage report. Default : 0.
* `-u,--unalign <index of first element>` Misalign source and destination buffers. Default : 0.
### Running multiple operations in parallel
@ -107,4 +108,4 @@ Note that the reported bandwidth is per group, hence to get the total bandwidth
## Copyright
NCCL tests are provided under the BSD license. All source code and accompanying documentation is copyright (c) 2016-2025, NVIDIA CORPORATION. All rights reserved.
NCCL tests are provided under the BSD license. All source code and accompanying documentation is copyright (c) 2016-2026, NVIDIA CORPORATION. All rights reserved.

View File

@ -1170,7 +1170,7 @@ int main(int argc, char* argv[], char **envp) {
"[-D,--device_implementation <implementation number> enable device implementation (default: 0, use NCCL implementation; requires -R 2 if > 0)] \n\t"
"[-V,--device_cta_count <number> set number of CTAs for device implementation (default: 16)] \n\t"
"[-M,--memory_report <0/1> enable memory usage report (default: 0)] \n\t"
"[-u,--unalign <index of first element>] \n\t"
"[-u,--unalign <index of first element> Misalign source and destination buffers (default: 0)] \n\t"
"[-h,--help]\n",
basename(argv[0]));
return 0;